Overview
Hangzhou Moxiang Big Data Technology is a Zhejiang University-linked AI company. Founder Dr. Gao Peng studied under academician Pan Yunhe. Described as China’s only technology company focused on banking-specific financial LLMs and large-scale pretrained AI employees. Jointly developed the Zhihai Jinpan financial model with Zhejiang University to serve China’s 2000+ banks. AI staff support voice services, human-machine collaboration, direct branch operations, remote banking, employee consultation, training, assistance and app avatars. Embedded MaaS and conversational agents assist marketing, customer service and recommendations. Zhihai Jinpan includes a deep-neural-network training/inference platform and pretrained algorithms and knowledge bases for 150 banking business categories. Its foundation supports hierarchical dialogue recognition, general and industry data/knowledge annotation and human-machine collaboration. Specialized product/scenario tools manage knowledge acquisition, retrieval and reasoning. Trained on a 1 billion-scale banking dataset, its AI employees include: 1. Direct-marketing assistant: capabilities comparable to a relationship manager with 1-3 years of experience, covering 150 businesses such as asset growth, credit cards and loans. Average conversion reportedly reaches 5%, or 2 times conventional slot-filling/search-based voice platforms. Integrates into CRM and work-tablet workflows for human-like telemarketing. 2. AIGC relationship-manager assistant: trained on selling points of hundreds of products for different audiences, analyzes customer situations and generates personalized marketing, wealth plans, benefits and scripts. 2.1. Knowledge questions: understands bank products, manuals and regulations to answer staff questions. 2.2. Personalized selling points: analyzes customers, suggests asset allocation and suitable products. 2.3. Script guidance: extracts strong personalized approaches from successful outbound conversations. 3. AIGC customer-service assistant: supports context summaries, issue classification, agent guidance, conversation summaries, ticket drafting and outbound strategies. Deployed at 100+ banks, with 8000 AI staff training and working at the six major banks, leading joint-stock banks, city commercial banks and rural commercial banks. Handles marketing, verification, reactivation, asset growth, installments and consumer-finance communications, improving staff efficiency and results. Holds 31 technology patents and over 200 software copyrights, plus national high-tech, provincial R&D-center and specialized innovative SME recognition. Established a joint financial-AI laboratory with Zhejiang University and owns core financial AI intellectual property. Included in EqualOcean’s 2020 top 100 potential AI commercialization companies, Hangzhou potential unicorns and the Kunpeng program. Jinpan, jointly released with Zhejiang University, has 7B and 13B models with further scaling possible. Retail-finance training sources include books, papers, knowledge graphs and dialogue texts. Targets efficient, intelligent and reliable financial questions and answers, text generation and knowledge reasoning. Core features address general models’ weak quantitative business problem-solving through retail-finance fine-tuning, specialist knowledge injection and external knowledge collaboration. Supports explainable provenance linking outputs to parameters, training data and external knowledge. Combines general model capabilities and financial generalization in moderate-sized models deployable privately with limited enterprise computing. Integrates local data and search components to strengthen output while meeting security and compliance requirements.
Connected profiles
Explore
More Finance products
Sources · 0Sources & provenance
English text is maintained separately from the original record. Consult the original source before citing a fact.
No source links are available for this profile.
Original source text
Names may use phonetic renderings when no English name is supplied. The original wording is retained below.
杭州摸象大数据科技有限公司是一家浙大系AI科技公司，创始人高鹏博士师从潘云鹤院士，是中国唯一专注于银行场景做金融垂直大语言模型（LLM）和大规模预训练AI员工的科技公司。和浙江大学联合研发大规模预训练的金融垂直大模型-智海金磐大模型，为中国的2000+银行提供高智能的AI员工，可应用于银行的智能语音、人机协同、网点直营、远程银行、员工咨询、培训和辅助、APP数字人等场景，嵌入在工作流程中，以MaaS平台和对话智能体的应用模式协助银行完成营销、客服和推荐服务。 【智海金磐】大语言模型（LLM）是垂直于银行行业的金融垂直大模型，内置基于深度神经网络的大模型训练&推理平台及预训练好的150个银行业务类目的深度算法模型和知识库。平台底座包括一套推理引擎用于分层对话识别，用于通用和行业层面的数据和知识标注、人机协同等智能体模块，也包括用于垂直和专业的产品和场景层的知识获取、检索、推理等知识库管理工具。 摸象科技基于垂直银行场景的10亿级高质量数据集训练，训练出的AI员工具有如下特点： 1、AI直营助手：具备1-3年客户经理的营销能力，可营销【资产提升、信用卡、贷款】等150类业务，目前平均营销转化率可达5%，成效是传统的基于填槽搜索式智能语音平台的2倍，可嵌入现有客户经理CRM系统或工作PAD的工作流中，帮助客户经理完成类似真人的电话营销，提升业绩； 2、AIGC客户经理助手：深度训练银行数百款金融产品针对不同用户群的卖点，可帮助客户经理实时分析每一个客户的营销战况，生成个性化营销策略、财富方案、权益建议和营销话术等。 2.1、知识随心问：可理解全行所有产品知识、操作手册、政策法规等，实现对客户经理有问必答； 2.2、个性化卖点策划：大模型可查询每一个客户并进行个性化分析，提出财富配置建议，配比推荐产品等； 2.3、营销话术指导：大模型筛选全行所有优质成功外呼对话，提炼出针对个性化场景的明星话术； 3、AIGC客服助手：面向银行客服中心坐席的AIGC的应用产品，可帮助客服坐席进行前情摘要、问题分类、座席指导、对话小结、工单撰写、外呼策略策划等，协助客服坐席降本增效。 目前摸象科技的AI员工解决方案已在100+银行中落地，有8000个AI员工在包括六大行、头部股份行和城农商行等单位训练和工作。AI员工对银行海量客户群体展开营销、核验、促活、资产提升、分期、消费金融等业务营销和服务沟通工作，大幅提升客户经理的工作效率和营销业绩，为银行数智化智能化运营提质降本增效！ 公司拥有国家级技术专利31项，200多项软件著作权，荣获国家高新技术企业、省级研发中心、专精特新中小企业等国家级资质认证，与浙江大学成立“浙大摸象金融人工智能联合实验室”，完全自有知识产权，具备垂直于金融行业的人工智能核心技术，被亿欧评选为2020年中国人工智能商业落地价值潜力企业100强，并入选杭州准独角兽企业、鲲鹏计划企业。【金磐大模型】是浙江大学和摸象科技联合发布的一个自主研发的垂直金融的语言大模型,目前模型规模7B、13B,可进一步扩展。训练的数据集垂直于零售金融方向,涵盖了金融书籍、论文,金融知识图谱、金融对话文本等多种数据源 【金磐大模型】的目标是为金融机构提供高效、智能、可信赖的语言服务,包括金融知识问答、金融文本生成、金融知识推理分析等多种应用场景。 【核心特征】解决通用大模型缺乏精准定量解决业务问题的能力,在金融零售领域实现精调适配、专业知识注入、外部知识库协同,以及解决模型输出内容和模型参数、训练数据和外源知识的可解释性溯源。同时具有大模型基础能力和金融领域泛化能力,模型体积小,参数量适中,单个金融企业有限算力可以承载,可以私有化部署,满足数据安全合规要求,同时模型能结合各种本地数据以及集成搜索组件以增强模型输出能力。